Accuracy of Regression-Based Imputation of Socioeconomic Status Components in PISA 2018

• In Event: Well-Being and Health

Abstract

In the Programme for International Student Assessment (PISA), students’ socio-economic status (SES) is measured with three components – family wealth, parents’ education, and parents’ occupation. When a student is missing data on only one component, a stochastic regression imputation is used to predict the missing component, and the imputed value is used to calculate the student’s SES score. Using PISA 2018 data from 76 countries, the study found that this imputation method was not very accurate in predicting the missing components of SES. It also found that the accuracy of the imputation method depended heavily on the strength of the relationship among the three components within a country, and that this relationship varied widely across countries.

Authors

• Selene Sunmin Lee, Educational Testing Service

• Nala Miller, Virginia Tech